Abstract
The use of molecular methods for detection of Chlamydia trachomatis is increasing in clinical laboratories. External quality assessment enables unbiased monitoring of the performance of laboratories in the detection of specific pathogens. This study details the results of molecular and enzyme immunosorbent assay (EIA) testing for C. trachomatis detection in simulated endocervical swab specimens recently distributed internationally by United Kingdom National External Quality Assessment Scheme for Microbiology (UK NEQAS for Microbiology) external quality assessment panels. The frequency of accurate detection of C. trachomatis in the panels ranged from 32 to 100%. Participants using molecular methods were significantly more likely to detect C. trachomatis in specimens than those using an EIA. Two strains were distributed with the panels: an L2 laboratory-adapted strain and an uncharacterized primary isolate. Further analysis indicated a difference in detection of C. trachomatis between specific methods only with the L2 strain at lower concentrations. In addition, eight negative specimens were distributed, and false positives were found to be rare by all methods included in the study.
Chlamydia trachomatis is an important sexually transmitted pathogen that causes upper and lower genital tract infections in males and females, leading to endometritis, infertility, and infections of the neonate. Clinical specimens that are tested for C. trachomatis include endocervical swabs, urethral swabs, urine specimens, conjunctival swabs, and vulvovaginal swabs. Traditional testing relied on cell culture and/or detection of C. trachomatis by enzyme immunosorbent assays (EIA) or immunofluorescence (IF). Due to the low sensitivity of culture, the new “gold standard,” known as the expanded gold standard, requires that a positive result be confirmed by repeat testing with a different assay (25, 28). Several molecular tests or nucleic acid amplification tests (NAAT), including BDProbeTecET, Gen-Probe AMP-CT, Gen-Probe APTIMA Combo 2, Roche AMPLICOR CT/NG, and Roche COBAS AMPLICOR CT/NG, are now commercially available; some are no longer available (Abbott LCx); and new technologies such as molecular beacons are being developed (1). Previous studies have attributed discrepant results with NAAT to inhibitors in the specimens, inappropriate specimen handling, cross-contamination, or poor sensitivity (4, 7, 10, 11, 16, 19, 23, 24). The reporting of false-positive results has medical, social, and psychological effects on the patient and results in increased health care costs and inappropriate treatment (6).
The United Kingdom National External Quality Assessment Scheme for Microbiology (UK NEQAS for Microbiology) has specialized in the production and performance analysis of external quality assessment (EQA) panels for a wide range of bacteria, fungi, parasites, and viruses for more than 30 years. An EQA panel for the detection of C. trachomatis by EIA and IF was introduced in 1991, and more-challenging specimens specifically designed to test NAAT methods have been included in the scheme from 2002. In addition to the EIA-IF schemes, a pilot scheme for the molecular detection of C. trachomatis is under development, and panels for testing by NAAT methods only have been distributed recently.
Verkooyen et al. (26) described the results obtained from an international EQA scheme consisting of freeze-dried urine specimens. The authors reported difficulty in the detection of C. trachomatis at lower concentrations and found no difference between the NAAT methods Roche AMPLICOR CT/NG, Roche COBAS AMPLICOR CT/NG, and Abbott LCx. Accurate detection of C. trachomatis was common, ranging from 89 to 100%. A further EQA study of detection of C. trachomatis in urine samples, based in Australia, found varying performance depending on the concentration (0 to 100%) and reported that low-level positives could not be detected consistently by a single test (18). In contrast to the results reported by Verkooyen et al. (26), Land et al. (18) found that at lower concentrations of C. trachomatis, the Roche AMPLICOR CT/NG assay was more sensitive than the Abbott LCx. However, Goessens et al. (14) examined the detection of C. trachomatis in female specimens (urine and swabs) and found no difference among three commercial tests (Abbott LCx, Gen-Probe AMP-CT, and Roche COBAS AMPLICOR CT/NG).
The aim of this retrospective study was to analyze the effective detection of C. trachomatis in simulated endocervical swab specimens by clinical laboratories participating in the EIA and molecular EQA schemes distributed internationally by UK NEQAS for Microbiology between January 2002 and June 2004. Negative specimens and those that were designed to challenge participants by including low levels of elementary bodies (EB) per milliliter were included in the study. Two C. trachomatis strains were included in the study: one L2 laboratory-adapted strain and one primary clinical isolate. Due to the large number of participants in the study, additional data were collected that provided insight into the differential performance of the methodologies in use in clinical laboratories. Therefore, the accurate detection of C. trachomatis by participants, different methods (EIA or NAAT), and commercial assays was compared with the C. trachomatis EQA specimens distributed. A total of eight negative specimens were also distributed, and the number of false positives reported was calculated.
MATERIALS AND METHODS
Specimen preparation.
Two strains of C. trachomatis were used in the study: a laboratory-adapted L2 strain (kindly donated by the Bristol Health Protection Agency [HPA]) cultured in McCoy cells (1 × 106 EB/ml) and a primary clinical isolate of unknown serotype, strain 6498, cultured in McCoy cells (3.4 × 106 EB/ml) from an endocervical sample from an 18-year-old female attending a clinic (kindly donated by Dudley Road Hospital, Birmingham, United Kingdom). Specific dilution series were made (see Table 1) in Earle's balanced salt solution (L2 strain) with 0.05% Bronidox (an antibacterial agent) or in 5% sucrose (strain 6498), and 0.5 ml was immediately dispensed into 2-ml plastic or glass vials. Specimens of the L2 strain were dispatched in liquid form, and those of 6498 in 5% sucrose were freeze-dried in a Lyoflex automated machine. The vacuum integrity of sealed freeze-dried vials was confirmed by spark testing followed by confirmation of a moisture content of <2% (Mitsubishi CA 100). Sealed freeze-dried vials were then stored at −30°C, and liquid vials were stored at 4°C, until distribution. Freeze-dried specimens were reconstituted in 0.5 ml of molecular-grade RNase-free water immediately prior to analysis in-house.
TABLE 1.
Performance of all methods taken together, EIA, and molecular methods
| Panel (no. of participants) | Date | Specimen | Resulta by:
|
P (EIA vs molecular methods)b | |||||
|---|---|---|---|---|---|---|---|---|---|
| All methods
|
EIA
|
Molecular methods
|
|||||||
| No. pos/total | % Pos (95% CI) | No. pos/total | % Pos (95% CI) | No. pos/total | % Pos (95% CI) | ||||
| L2 strain CT001 | |||||||||
| 1500 (289) | Jan. 2002 | 6157 | 116/267 | 43.4 (37.4-49.6) | 14/162 | 8.6 (4.8-14.0) | 102/105 | 97.1 (91.9-99.4) | <0.0001 |
| 1523 (291) | Mar. 2002 | 6233 | 117/279 | 41.9 (36.1-47.9) | 8/159 | 5.0 (2.2-9.7) | 109/120 | 90.8 (84.2-95.3) | <0.0001 |
| 1565 (294) | Jul. 2002 | 6346 | 112/292 | 38.4 (32.9-44.1) | 6/156 | 3.9 (1.4-8.2) | 106/136 | 77.9 (70.0-84.6) | <0.0001 |
| 1585 (293) | Oct. 2002 | 6424 | 128/290 | 44.1 (38.5-49.9) | 6/152 | 3.9 (1.5-8.4) | 122/138 | 88.4 (81.9-93.2) | <0.0001 |
| 1614 (295) | Mar. 2003 | 6547 | 102/284 | 35.9 (30.5-41.7) | 8/150 | 5.3 (2.3-10.3) | 94/134 | 70.1 (61.6-77.7) | <0.0001 |
| 1660 (285) | June 2003 | 6680 | 240/287 | 83.6 (79.1-87.5) | 101/146 | 69.2 (61.0-76.6) | 139/141 | 98.6 (94.9-99.8) | <0.0001 |
| 6681 | 111/283 | 39.2 (33.7-45.0) | 7/147 | 4.8 (1.9-9.6) | 104/136 | 76.5 (68.4-83.3) | <0.0001 | ||
| 1689 (280) | Oct. 2003 | 6788 | 200/254 | 78.7 (73.5-83.2) | 61/112 | 54.5 (44.8-63.9) | 139/142 | 97.9 (93.9-99.6) | <0.0001 |
| 1733 (280) | Mar. 2004 | 6962 | 80/248 | 32.3 (26.7-38.3) | 3/132 | 2.3 (0.5-6.5) | 77/116 | 66.4 (57.0-74.9) | <0.0001 |
| Clinical isolate 6498 | |||||||||
| 1734 (91) | Mar. 2004 | 6963 | 84/84 | 100 (96.5-100) | NA | NA | 84/84 | 100 (96.5-100) | NA |
| 6964 | 84/84 | 100 (96.5-100) | NA | NA | 84/84 | 100 (96.5-100) | NA | ||
| 1785 (103) | June 2004 | 7112 | 98/98 | 100 (96.9-100) | NA | NA | 98/98 | 100 (96.9-100) | NA |
| 7113 | 32/98 | 32.7 (23.5-42.9) | NA | NA | 32/98 | 32.7 (23.5-42.9) | NA | ||
| 7115 | 98/98 | 100 (96.9-100) | NA | NA | 98/98 | 100 (96.9-100) | NA | ||
| Negative | |||||||||
| 1500 (289) | Jan. 2002 | 6158 | 3/269 | 1.1 (0.3-3.2) | 2/164 | 1.2 (0.2-4.3) | 1/105 | 1.0 (0.02-5.2) | 1.0 |
| 1523 (291) | Mar. 2002 | 6235 | 4/271 | 1.5 (0.4-3.7) | 3/151 | 2.0 (0.4-5.7) | 1/120 | 0.8 (0.02-4.6) | 0.6 |
| 1565 (294) | July 2002 | 6347 | 1/300 | 0.3 (0.01-1.8) | 1/164 | 0.6 (0.02-3.4) | 0/136 | 0 (0-2.2) | 1.0 |
| 1585 (293) | Oct. 2002 | 6423 | 1/290 | 0.3 (0.01-1.9) | 1/152 | 0.6 (0.02-3.6) | 0/138 | 0 (0-2.2) | 1.0 |
| 1614 (295) | Mar. 2003 | 6545 | 4/291 | 1.4 (0.4-3.4) | 3/151 | 2.0 (0.4-5.7) | 1/140 | 0.7 (0.02-3.9) | 0.6 |
| 1660 (285) | June 2003 | 6682 | 3/281 | 1.1 (0.2-3.0) | 2/145 | 1.4 (0.2-4.9) | 1/136 | 0.7 (0.05-4.0) | 1.0 |
| 1733 (280) | Mar. 2004 | 6959 | 3/283 | 1.1 (0.3-3.1) | 1/135 | 0.7 (0.02-4.1) | 2/148 | 1.4 (0.2-4.8) | 1.0 |
| 1785 (103) | June 2004 | 7114 | 1/98 | 1.0 (0.03-5.6) | NA | NA | 1/98 | 1.0 (0.03-5.6) | NA |
pos, positive and equivocal; total, total number of tests (some participants used more than one method). NA, not applicable.
By Fisher's exact text.
A sample specimen from each dilution was sterility tested by inoculation of 100 μl of the liquid or pre-freeze-dried specimen into 5 ml of nutrient broth followed by incubation for 48 h at room temperature and +37°C. No specimens yielded microbiological growth after 48 h of incubation. Specimens were monitored for stability at a range of temperatures; liquid specimens were found to be stable at 4°C for at least 2 months, and freeze-dried specimens were found to be stable at 4°C, room temperature, and −30°C for as long as 12 months (data not shown).
Before distribution, the specimens were tested for C. trachomatis by a range of laboratories using BDProbeTecET, Gen-Probe AMP-CT, in-house PCR, and Roche AMPLICOR CT/NG (for all panels) and Dade Behring Microtrak II EIA and Dako IDEIA PCE (for panels 1500 to 1733).
To monitor specimen delivery and the effects of transit on the specimen, five specimen sets from each panel were posted within each consignment to five different sites around the United Kingdom and, upon receipt, were returned by post to UK NEQAS for Microbiology for retesting.
Distribution information.
Each proficiency panel consisted of four specimens, a reply form, and an instruction sheet. Participants were provided with instructions to reconstitute freeze-dried specimens in 0.5 ml of molecular-grade RNase-free water immediately prior to analysis, add 0.1 ml of the reconstituted specimen to their transport medium, and extract by the usual laboratory routine. For liquid specimens, participants were instructed to examine 0.1 ml of each specimen for C. trachomatis by their routine method. Participants were requested to test for the presence of C. trachomatis in each specimen and report results within 21 days of posting. Results were received by fax, post, or e-mail or via a World Wide Web online reporting system. Ten panels have been distributed since 2002, and the data presented here include a total of 22 specimens, 14 positive and 8 negative, from these distributions (Table 1). The median concentration of C. trachomatis in each specimen, in EB per milliliter, is shown in Table 2. Specimens were distributed to 91 to 295 participating laboratories (Table 1), predominantly in the United Kingdom, but also in other countries including Austria, Belgium, Denmark, Finland, Greece, Hong Kong, Ireland, Israel, Italy, Kuwait, The Netherlands, Norway, Portugal, Slovenia, Sweden, Switzerland, and Zambia.
TABLE 2.
Performance of molecular methods
| Specimen | Dilution (EB/mla) | Resultb by:
|
Risk difference coefficient (P) for BDProbeTecET vs Roche AMPLICOR | |||||
|---|---|---|---|---|---|---|---|---|
| Abbott LCx
|
BDProbeTecET
|
Roche AMPLICOR/COBASCT/NG
|
||||||
| No. pos/total | % Pos (95% CI) | No. pos/total | % Pos (95% CI) | No. Pos/total | % Pos (95% CI) | |||
| L2 strain CT001 | ||||||||
| 6680 | 1:200 (5,000) | 12/12 | 100 (76.0-100) | 59/59 | 100 (93.9-100) | 66/66 | 100 (94.5-100) | NA |
| 6788 | 1:500 (2,000) | 4/4 | 100 (51.0-100) | 65/65 | 100 (94.4-100) | 66/67 | 98.5 (92.0-99.7) | −0.15 (0.31) |
| 6157 | 1:1,000 (1,000) | 29/29 | 100 (88.3-100) | 30/32 | 93.8 (79.9-98.3) | 39/39 | 100 (91.0-100) | 0.63 (0.14) |
| 6962 | 1:1,200 (833) | 2/2 | 100 (34.2-100) | 11/35 | 31.4 (18.6-47.9) | 60/71 | 84.5 (74.4-91.1) | 0.53 (0.00) |
| 6424 | 1:1,500 (667) | 28/30 | 93.3 (78.7-98.2) | 40/51 | 78.4 (65.4-87.5) | 52/53 | 98.1 (90.1-99.7) | 0.19 (0.001) |
| 6233 | 1:2,000 (500) | 30/31 | 96.8 (83.8-99.4) | 32/40 | 80.0 (65.2-89.5) | 42/44 | 95.5 (84.9-98.7) | 0.26 (0.001) |
| 6681 | 1:2,000 (500) | 10/12 | 83.3 (55.2-95.3) | 36/57 | 63.2 (50.2-74.5) | 56/63 | 88.9 (78.8-94.5) | 0.15 (0.03) |
| 6346 | 1:2,500 (400) | 27/33 | 81.8 (65.6-91.4) | 26/47 | 55.3 (41.3-68.6) | 51/52 | 98.1 (89.9-99.7) | 0.43 (0.00) |
| 6547 | 1:3,000 (333) | 8/11 | 72.7 (43.4-90.3) | 27/59 | 45.8 (33.7-58.3) | 54/58 | 93.1 (83.6-97.3) | 0.47 (0.00) |
| Total for L2 | NA | 150/164 | 91.5 (86.1-95.3) | 326/445 | 73.3 (68.9-77.3) | 486/613 | 94.7 (92.4-96.5) | NA |
| Clinical isolate 6498 | ||||||||
| 6963 | 1:40 (85,750) | NA | NA | 47/47 | 100 (93.8-100) | 35/35 | 100 (91.8-100) | NA |
| 6964 | 1:400 (8,575) | NA | NA | 47/47 | 100 (93.8-100) | 35/35 | 100 (91.8-100) | 1.4 × 10−24 (1.0) |
| 7112 | 1:400 (8,575) | NA | NA | 54/54 | 100 (94.6-100) | 41/41 | 100 (92.9-100) | 1.3 × 10−15 (1.0) |
| 7113 | 1:250,000 (14) | NA | NA | 18.54 | 33.3 (21.1-47.5) | 13/41 | 31.7 (18.1-48.1) | −0.16 (0.87) |
| 7115 | 1:1,200 (2.858) | NA | NA | 54/54 | 100 (94.6-100) | 41/41 | 100 (92.9-100) | 4.4 × 10−15 (1.0) |
| Total for 6498 | NA | NA | NA | 220/256 | 85.9 (81.3-89.7) | 165/193 | 85.5 (80.1-89.8) | NA |
| Negative | ||||||||
| 6158 | NA | 1/29 | 3.5 (0.1-17.7) | 0/32 | 0 (0-8.9) | 0/39 | 0 (0-7.4) | NA |
| 6235 | NA | 0/32 | 0 (0-8.9) | 0/39 | 0 (0-7.4) | 1/44 | 2.3 (0.06-12.0) | NA |
| 6347 | NA | 0/33 | 0 (0-8.7) | 0/47 | 0 (0-6.2) | 0/52 | 0 (0-5.6) | NA |
| 6423 | NA | 0/30 | 0 (0-9.5) | 0/51 | 0 (0-5.7) | 0/53 | 0 (0-5.7) | NA |
| 6545 | NA | 0/11 | 0 (0-23.8) | 0/61 | 0 (0-4.8) | 1/62 | 1.6 (0.04-8.7) | NA |
| 6682 | NA | 0/12 | 0 (0-22.1) | 1/58 | 1.7 (0.04-9.2) | 0/62 | 0 (0-4.7) | NA |
| 6959 | NA | 0/2 | 0 (0-77.6) | 1/66 | 1.5 (0.04-8.2) | 0/72 | 0 (0-4.1) | NA |
| 7114 | NA | NA | NA | 0/53 | 0 (0-5.5) | 1/42 | 2.4 (0.06-12.6) | NA |
Estimated EB per milliliter are extrapolated from IF data for a 1:1,000 dilution (4 × 10-μl wells).
NA, not applicable.
Data analysis.
The number of participants detecting C. trachomatis was calculated in total and by method per specimen. The EIA and molecular method groups were compared by using Fisher's exact test. The methods included in the category of molecular tests (Table 1) were as follows: Abbott LCx, BDProbeTecET, Gen-Probe PACE 2, Gen-Probe AMP-CT, in-house PCR (amplifying a 207-bp region of a cryptic plasmid), and Roche AMPLICOR CT/NG. The methods included in the EIA category (Table 1) were as follows: Abbott Chlamydiazyme IMx, Beckman Coulter Access Chlamydia, bioMérieux VIDAS, Bio-Rad Pathfinder EIA, Murex Diagnostics Inc. BioStar Chlamydia OIA, Dade Behring MicroTrak Chlamydia EIA, Dako IDIEA, Dako IDEIA PCE, Launch Diagnostics Phadebact Chlamydia EIA, and Unipath Clearview Chlamydia. Result collation is dependent on the information provided by participants, and the majority of users of Roche methods stated “Roche” only and did not specify Roche AMPLICOR CT/NG or Roche COBAS AMPLICOR CT/NG. For the purposes of this study, the two methods have been collated into a single group, termed Roche AMPLICOR/COBAS CT/NG. The detection of C. trachomatis by users of the Roche AMPLICOR/COBAS CT/NG method group was compared to detection by the BDProbeTecET method by using risk difference analysis (STATA statistical analysis package) with differences calculated at each level of dilution. Similar analysis was performed to compare the detection of the L2 strain by the Abbott LCx method to that by the Roche AMPLICOR/COBAS CT/NG method group and the BDProbeTecET method in specimens distributed when the Abbott LCx method was still available.
RESULTS
Results are summarized in Tables 1 and 2. Overall detection of C. trachomatis by participants ranged from 32 to 100%.
For the L2 strain, the rate of detection of C. trachomatis by participants ranged from 32.3 to 83.6% (Table 1). The rate of C. trachomatis detection by users of molecular methods ranged from 66.4 to 98.6% and was significantly higher than that with EIA methods, which ranged from 2.3 to 69.2% (P = 0.00 by Fisher's exact test [Table 1]).
When all data for the L2 strain were compared, no difference was noted between the Abbott LCx, the BDProbeTecET (risk difference coefficient, 7.93 × 10−14; P = 1.0), and the Roche AMPLICOR/COBAS CT/NG method group (risk difference coefficient, 2.89 × 10−14; P = 1.0). There was no significant difference between molecular methods at dilutions of 1:200 through 1:2,000. However, detection of C. trachomatis by all methods showed significant differences at dilutions greater than 1:2,000 (risk difference coefficients, −1.81 at 1:2,500 [P = 0.007] and −0.27 at 1:3,000 [P = 0.04]), indicating that all methods are less likely to detect C. trachomatis at lower concentrations of the pathogen. The Abbott LCx accurately detected C. trachomatis in 72.7 to 100% of specimens, the BDProbeTecET method detected it in 31.4 to 100% of specimens, and the Roche AMPLICOR/COBAS CT/NG method group detected it in 84.5 to 100% of specimens. Significant differences were found in the likelihood that each method would report a positive result when risk difference analysis was used to compare the Abbott LCx, the BDProbeTecET method, and the Roche AMPLICOR/COBAS CT/NG method group, with differences calculated at each level of dilution. The BDProbeTecET method was less likely to give a positive result at dilutions of 1:1,200 and higher (risk difference coefficients, −0.69 at 1:1,200 [P = 0.00], −0.14 at 1:1,500 [P = 0.04], −0.2 at 1:2,000 [P = 0.11], −0.17 at 1:2,000 [P = 0.02], −0.26 at 1:2,500 [P = 0.007], and −0.27 at 1:3,000 [P = 0.071]), and the Abbott LCx was less likely to give a positive result at dilutions of 1:2,500 and higher (risk difference coefficients, 0.16 at 1:2,500 [P = 0.02] and 0.2 at 1:3,000 [P = 0.14]).
To enable comparison of results for the L2 strain and strain 6498 (for which no Abbott LCx data were available), further analysis was performed to compare all L2 strain data for the BDProbeTecET and the Roche AMPLICOR/COBAS CT/NG method group. No difference was found when all data were compared (risk difference coefficient, 2.03 × 10−14, P = 1.0). Significant differences were found in the likelihood that each method would report a positive result when risk difference analysis was used to compare the BDProbeTecET method and the Roche AMPLICOR/COBAS CT/NG method group, with differences calculated at each level of dilution (Table 2). The BDProbeTecET method was less likely to give a positive result than the Roche AMPLICOR/COBAS CT/NG method group at dilutions of 1:1,200 to 1:3,000 (Table 2) for the L2 strain.
The rate of detection of strain 6498 ranged from 32.7 to 100% for the specimens distributed (Table 1). Only molecular tests were performed on these specimens. No difference was noted between the BDProbeTecET and the Roche AMPLICOR/COBAS CT/NG method group when all data for the specimens containing strain 6498 were compared (risk difference coefficient, −5.02 × 10−24, P = 1.0). In addition, comparison of C. trachomatis detection by the two methods between the 1:40 dilution and other dilutions gave a significant difference at the 1:250,000 dilution only (risk difference coefficient at 1:250,000, −0.67; P = 0.00), indicating that both methods are less likely to detect C. trachomatis at lower concentrations of the bacterium. The BDProbeTecET method accurately detected C. trachomatis in 33.3 to 100% of specimens, and the Roche AMPLICOR/COBAS CT/NG method group detected it in 31.7 to 100%. Unlike the findings for the L2 strain, no significant difference was found in the likelihood that each method would report a positive result when risk difference analysis was used to compare the BDProbeTecET method and the Roche AMPLICOR/COBAS CT/NG method group with differences calculated at each level of dilution (Table 2). The two methods were equally likely to give a positive result with strain 6498.
There was no significant difference in the detection of the two different C. trachomatis strains by users of all molecular tests (P = 0.76 by Fisher's exact test; for strain L2, 992 of 1,168 tests [84.9%] were positive [95% confidence interval {95% CI}, 82.8 to 86.9%]; for strain 6498, 396 of 462 tests [85.7%] were positive [95% CI, 82.2 to 88.8%]). Similarly, comparison of the detection of the two C. trachomatis strains by users of the BDProbeTecET method showed no difference (P = 0.18 by Fisher's exact test), and no difference was noted in the detection of the two strains by the Roche AMPLICOR/COBAS CT/NG method group tests (P < 0.42 by Fisher's exact test). However, comparison of BDProbeTecET with the Roche AMPLICOR/COBAS CT/NG method group showed that the latter was more likely to detect C. trachomatis in specimens containing the L2 strain (P < 0.008 by Fisher's exact test), whereas no difference was seen in the detection of strain 6498 (P = 1.0 by Fisher's exact test).
For the eight negative specimens that were distributed, 0.3 to 1.5% of results were positive. A total of 20 positive results and 2,080 negative results were received. The positive results were received from 18 different participants using a range of methods (bioMérieux VIDAS [n = 4], Murex Diagnostics [n = 1], Dako IDIEA [n = 1], Launch Diagnostics Phadebact Chlamydia EIA [n = 3], Unipath Clearview Chlamydia [n = 4], Abbott LCx [n = 1], BDProbeTecET [n = 3], Gen-Probe AMP-CT [n = 1], and Roche AMPLICOR CT/NG [n = 2]). A single participant reported three negative specimens as positive, whereas all other false positives were received from separate participants.
DISCUSSION
EQA enables monitoring of the performance of laboratories with defined specimens from a position external to the host organization and therefore in an unbiased fashion. Such monitoring enables the detection of specific pathogens to be compared among laboratories both nationally and internationally depending on the participant base, highlighting areas for improvement and promoting continuing maintenance and improvements in clinical diagnostic standards. Several aspects of routine procedures can be tested by EQA, such as issues of specimen handling, test performance, result reporting, and general laboratory performance over time. However, some aspects, such as the individual routines specific to laboratories (18), routine reporting using internal standards, and individual laboratory workers, cannot be monitored. Nonetheless, EQA provides valuable insight into the standard of clinical diagnostic testing by laboratories. In addition to monitoring standards, the process of EQA collates information that can be utilized to provide insight into the performance of specific laboratory methods (18, 26). In this study, EQA enabled the comparison of C. trachomatis detection by differing assay types (EIA and NAAT) and also the comparison of the three main NAAT used by participants in the study (Abbott LCx, BDProbeTecET, and the Roche AMPLICOR/COBAS CT/NG method group), with two distinct sources of C. trachomatis.
The ability of laboratories to detect C. trachomatis in the EQA panels included in this study has been variable. In general, those using EIAs were significantly less likely to report detection of C. trachomatis than those using the molecule-based NAAT. With a specimen containing approximately 833 EB/ml, only 2.3% of laboratories using EIAs detected C. trachomatis, in comparison to 66.4% of those using NAAT. These findings are in concordance with those of other studies, and it is now accepted that NAAT are superior to EIA-based tests for detection of C. trachomatis (22, 28). However, in this study only those specimens that were designed to be challenging to participants were included, and at higher concentrations such a difference may not exist. Furthermore, NAAT are not without problems, including the effects of inhibitors in the specimens, contamination, sensitivity, reproducibility, and variation in the efficiency of nucleic acid extraction (4, 6, 7, 10, 11, 15, 16, 19, 23, 24).
At higher concentrations of C. trachomatis, detection by NAAT was excellent, irrespective of the method used or the strain under test. Interestingly, we found that the BDProbeTecET method was less likely to detect C. trachomatis than the Roche AMPLICOR/COBAS CT/NG method group at lower concentrations of the L2 strain only. With the clinical isolate no difference was seen, a finding similar to the results of Chan et al. (8). Further analysis indicated that differences in strain detection did not occur with the BDProbeTecET method or with the Roche AMPLICOR/COBAS CT/NG method group. To date at least 18 serotypes of C. trachomatis have been recognized (12, 21, 27). Serotypes are distinguished by differences in the sequence of the ompA gene, which encodes the major outer membrane protein (MOMP). Although MOMP types differ by geographical region, no differences are seen between the serotypes present in males and females in the same population (21). Specific serotypes are associated with particular infections: types A, B, and Ba are commonly associated with trachoma, types C to K are associated with lower and upper genital tract infections, and the lymphogranuloma venereum serotypes L1, L2, L2a, and L3 are associated with more-severe infections such as proctitis and lymphadenitis (2, 5, 21). Furthermore, a strain has been associated with concurrent infection with Neisseria gonorrhoeae (3). Although the serotype of the clinical isolate used in this study, 6498, is not known, it is possible that differences in performance could be attributed to the differential efficacy of the Roche AMPLICOR/COBAS CT/NG method group with different strain types. However, the product insert for the Roche AMPLICOR/COBAS CT/NG method group states that the lower detection limit is the same for all 12 serotypes tested: 80 inclusion-forming units/ml. It is possible that strain 6498 is also an L2 strain, although this would be unusual for a clinical isolate from the United Kingdom. In the future, it would be interesting to distribute defined serotypes at known concentrations in EQA panels to determine whether differences in detection and/or sensitivity actually occur with different serotypes in clinical laboratories. It is also possible that these results have highlighted the fact that the BDProbeTecET method is less able to detect the L2 strain at lower concentrations, and analysis comparing the Abbott LCx results to those of the BDProbeTecET method and the Roche AMPLICOR/COBAS CT/NG method group showed that the BDProbeTecET method was less likely to detect the L2 strain in the lower-concentration specimens whereas the Roche AMPLICOR/COBAS CT/NG method group was more likely to detect the L2 strain. Unfortunately, due to discontinuation of the manufacture of the Abbott LCx test, comparison with strain 6498 was not possible. Interestingly, McCartney et al. (20) found that the sensitivity and specificity of the BDProbeTecET method were comparable to those of the Abbott LCx in testing of male urine and (female) endocervical samples; however, a lower detection rate was found with the BDProbeTecET method for female urine specimens, which contain lower concentrations of C. trachomatis. Furthermore, the product insert information for the BDProbeTecET method states that detection levels differ depending on the serotype in question, ranging from 5 to 200 EB/reaction, and that 15 EB/reaction is the limit of detection for type LGV2. Differences in the sensitivity of detection may affect epidemiological data, and most importantly, the identification of infected patients may be missed, resulting in inappropriate treatment or no treatment and the development of serious physical sequelae, psychological effects, and further transmission of the pathogen.
The rate of false positives in this study was low (0.3 to 1.5%) and was not associated with any method. False positives were common in a single participating laboratory only, which reported three consecutive negative specimens as positive; all other false positives were reported by separate participants. The results found in this study are similar to those reported by Land et al. (0 to 3% false-positive C. trachomatis results with urine specimens [18]).
A major drawback of this study is the use of simulated specimens in place of real clinical specimens. However, it would be very difficult to obtain the required volume of endocervical swab medium with sufficient uniformity of consistency and of C. trachomatis concentrations. Land et al. (18) reported that clinical urine specimens are problematic when used in EQA due to issues of stability and storage, and Kellogg et al. (17) noted that differences in detection are highly dependent on the quality of endocervical specimens. The benefits of using simulated specimens include the ability to define stability, concentration, and uniformity. The major difference between the two strains used in this study lies in the preparation of specimens with the chemical constituents of the medium used and the physical state of the specimens (liquid or freeze-dried). It is possible that these physical and chemical differences could have affected extraction and amplification efficiency, resulting in the differences in performance with the L2 strain. Strain 6498 was freeze-dried and was made with a liquid matrix different from that for the L2 strain. The process of freeze-drying could have damaged the C. trachomatis cells by disrupting cellular integrity, thereby releasing nucleic acid more readily. Freeze-drying could also have had an effect on the aggregation of the bacterial cells. However, the latter possibility is unlikely, because upon IF analysis, no aggregates were seen in freeze-dried specimens (data not shown). It is possible that the composition of the matrix could affect the detection of C. trachomatis, and specific chemical inhibitors and inhibitors in urine are known (4, 9, 19, 23). However, this study did not include inhibitors or their indicators such as hemoglobin, nitrates, and β-human chorionic gonadotropin in the specimens, although Earle's balanced salt solution with Bronidox and sucrose were used. Land et al. (18) included some specimens with inhibitors and noted that a significant number of participants reported a positive specimen with inhibitors as negative; these differences were not linked to specific methods.
A variety of sample types are tested for C. trachomatis in clinical laboratories. A recent poll of participants by UK NEQAS showed that 100% routinely test endocervical swabs and urine samples by molecular methods, with endocervical swabs representing the most frequently tested sample type. Of these laboratories, 51% also test other sample types (throat, rectal, and conjunctival swabs) less frequently. Varying the specimen types for EQA to include both endocervical swabs and urine specimens would more accurately reflect procedures in the clinical laboratory.
The current expanded gold standard for C. trachomatis detection includes the confirmation of a positive result with a second, different test (28). In 2002, meta-analysis showed that the majority of laboratories use one test only and do not confirm with a second, distinct test (28). In a recent UK NEQAS for Microbiology scheme (distribution 1734), 78% of participants reported that they routinely confirm positive results for C. trachomatis whereas 7% reported that they do not routinely confirm positives. Of those confirming positive results, only 1% used a different NAAT while 72% confirmed positive results by repeat testing with the same assay. Gaydos et al. (13) expressed concern at the cost of repeat testing. The retention of two separate NAAT platforms may simply be beyond the financial and staff resources of smaller laboratories.
In summary, the performance of participating laboratories in the detection of C. trachomatis in simulated EQA specimens was diverse. The detection of C. trachomatis by participants using NAAT was superior to that by those using EIAs. False positives were rare, and detection was excellent, at higher concentrations of C. trachomatis irrespective of the method used or the strain under test. The Roche AMPLICOR/COBAS CT/NG method group was more likely to detect the L2 strain than the Abbott LCx and BDProbeTecET methods in specimens containing lower concentrations, whereas no difference was noted between the Roche AMPLICOR/COBAS CT/NG method group and the BDProbeTecET method with clinical isolate 6498. The use of simulated specimens enabled powerful use of EQA results by testing differing strains and concentrations, reflecting the concentrations found in varying specimen types. The chemical matrix, physical presentation, and strain used in EQA specimens may affect results and should be considered in the design of EQA specimens. Future EQA studies should include a range of specimen types and a variety of C. trachomatis subtypes so as to more accurately reflect routine procedures in clinical diagnostic laboratories.
Acknowledgments
UK NEQAS for Microbiology thanks HPA Bristol for the L2 strain; Dudley Road Hospital for strain 6498; Harry Mallinson, University Hospital Aintree, Liverpool, United Kingdom, for critical reading of the manuscript; HPA Bristol, Noble's Hospital Douglas, St. Mary's Hospital London, University College Hospital London, and University Hospital Aintree for providing predistribution testing; Kathleen Baster, HPA Centre for Infections, CDSC Colindale, for statistical advice; and the UK NEQAS for Microbiology participants, without whom this study would not have been possible.
REFERENCES
- 1.Abravaya, K., J. Huff, R. Marshall, B. Merchant, C. Mullen, G. Schneider, and J. Robinson. 2003. Molecular beacons as diagnostic tools: technology and applications. Clin. Chem. Lab. Med. 41:468-474. [DOI] [PubMed] [Google Scholar]
- 2.Barnes, R. C., A. M. Rompalo, and W. E. Stamm. 1987. Comparison of Chlamydia trachomatis serovars causing rectal and cervical infections. J. Infect. Dis. 156:953-958. [DOI] [PubMed] [Google Scholar]
- 3.Batteiger, B. E., J. Fraiz, W. J. Newhall, B. P. Katz, and R. B. Jones. 1989. Association of recurrent chlamydial infection with gonorrhea. J. Infect. Dis. 159:661-669. [DOI] [PubMed] [Google Scholar]
- 4.Berg, E. S., G. Anestad, H. Moi, G. Storvold, and K. Skaug. 1997. False-negative results of a ligase chain reaction assay to detect Chlamydia trachomatis due to inhibitors in urine. Eur. J. Clin. Microbiol. Infect. Dis. 16:727-731. [DOI] [PubMed] [Google Scholar]
- 5.Boisvert, J. F., L. A. Koutsky, R. J. Suchland, and W. E. Stamm. 1999. Clinical features of Chlamydia trachomatis rectal infection by serovar among homosexually active men. Sex. Transm. Dis. 26:392-398. [DOI] [PubMed] [Google Scholar]
- 6.Boyadzhyan, B., T. Yashina, J. H. Yatabe, M. Patnaik, and C. S. Hill. 2004. Comparison of the APTIMA CT and GC assays with the APTIMA combo 2 assay, the Abbott LCx assay, and direct fluorescent-antibody and culture assays for detection of Chlamydia trachomatis and Neisseria gonorrhoeae. J. Clin. Microbiol. 42:3089-3093. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Castriciano, S., K. Luinstra, D. Jang, J. Patel, J. Mahony, J. Kapala, and M. Chernesky. 2002. Accuracy of results obtained by performing a second ligase chain reaction assay and PCR analysis on urine samples with positive or near-cutoff results in the LCx test for Chlamydia trachomatis. J. Clin. Microbiol. 40:2632-2634. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Chan, E. L., K. Brandt, K. Olienus, N. Antonishyn, and G. B. Horsman. 2000. Performance characteristics of the Becton Dickinson ProbeTec System for direct detection of Chlamydia trachomatis and Neisseria gonorrhoeae in male and female urine specimens in comparison with the Roche Cobas System. Arch. Pathol. Lab. Med. 124:1649-1652. [DOI] [PubMed] [Google Scholar]
- 9.Chernesky, M., D. Jang, S. Chong, J. Sellors, and J. Mahony. 2003. Impact of urine collection order on the ability of assays to identify Chlamydia trachomatis infections in men. Sex. Transm. Dis. 30:345-347. [DOI] [PubMed] [Google Scholar]
- 10.Chernesky, M. A., D. Jang, J. Sellors, K. Luinstra, S. Chong, S. Castriciano, and J. B. Mahony. 1997. Urinary inhibitors of polymerase chain reaction and ligase chain reaction and testing of multiple specimens may contribute to lower assay sensitivities for diagnosing Chlamydia trachomatis infected women. Mol. Cell. Probes 11:243-249. [DOI] [PubMed] [Google Scholar]
- 11.Chong, S., D. Jang, X. Song, J. Mahony, A. Petrich, P. Barriga, and M. Chernesky. 2003. Specimen processing and concentration of Chlamydia trachomatis added can influence false-negative rates in the LCx assay but not in the APTIMA Combo 2 assay when testing for inhibitors. J. Clin. Microbiol. 41:778-782. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Dean, D., R. J. Suchland, and W. E. Stamm. 2000. Evidence for long-term cervical persistence of Chlamydia trachomatis by omp1 genotyping. J. Infect. Dis. 182:909-916. [DOI] [PubMed] [Google Scholar]
- 13.Gaydos, C. A., M. Theodore, N. Dalesio, B. J. Wood, and T. C. Quinn. 2004. Comparison of three nucleic acid amplification tests for detection of Chlamydia trachomatis in urine specimens. J. Clin. Microbiol. 42:3041-3045. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Goessens, W. H., J. W. Mouton, W. I. van der Meijden, S. Deelen, T. H. van Rijsoort-Vos, N. Lemmens-den Toom, H. A. Verbrugh, and R. P. Verkooyen. 1997. Comparison of three commercially available amplification assays, AMP CT, LCx, and COBAS AMPLICOR, for detection of Chlamydia trachomatis in first-void urine. J. Clin. Microbiol. 35:2628-2633. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Gronowski, A. M., S. Copper, D. Baorto, and P. R. Murray. 2000. Reproducibility problems with the Abbott Laboratories LCx assay for Chlamydia trachomatis and Neisseria gonorrhoeae. J. Clin. Microbiol. 38:2416-2418. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Jensen, I. P., P. Thorsen, and B. R. Moller. 1997. Sensitivity of ligase chain reaction assay of urine from pregnant women for Chlamydia trachomatis. Lancet 349:329-330. [DOI] [PubMed] [Google Scholar]
- 17.Kellogg, J. A., J. W. Seiple, J. L. Klinedinst, and J. S. Levisky. 1991. Impact of endocervical specimen quality on apparent prevalence of Chlamydia trachomatis infections diagnosed using an enzyme-linked immunosorbent assay method. Arch. Pathol. Lab. Med. 115:1223-1227. [PubMed] [Google Scholar]
- 18.Land, S., S. Tabrizi, A. Gust, E. Johnson, S. Garland, and E. M. Dax. 2002. External quality assessment program for Chlamydia trachomatis diagnostic testing by nucleic acid amplification assays. J. Clin. Microbiol. 40:2893-2896. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Mahony, J., S. Chong, D. Jang, K. Luinstra, M. Faught, D. Dalby, J. Sellors, and M. Chernesky. 1998. Urine specimens from pregnant and nonpregnant women inhibitory to amplification of Chlamydia trachomatis nucleic acid by PCR, ligase chain reaction, and transcription-mediated amplification: identification of urinary substances associated with inhibition and removal of inhibitory activity. J. Clin. Microbiol. 36:3122-3126. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.McCartney, R. A., J. Walker, and A. Scoular. 2001. Detection of Chlamydia trachomatis in genitourinary medicine clinic attendees: comparison of strand displacement amplification and ligase chain reaction. Br. J. Biomed. Sci. 58:235-238. [PubMed] [Google Scholar]
- 21.Millman, K., C. M. Black, R. E. Johnson, W. E. Stamm, R. B. Jones, E. W. Hook, D. H. Martin, G. Bolan, S. Tavare, and D. Dean. 2004. Population-based genetic and evolutionary analysis of Chlamydia trachomatis urogenital strain variation in the United States. J. Bacteriol. 186:2457-2465. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Newhall, W. J., R. E. Johnson, S. DeLisle, D. Fine, A. Hadgu, B. Matsuda, D. Osmond, J. Campbell, and W. E. Stamm. 1999. Head-to-head evaluation of five chlamydia tests relative to a quality-assured culture standard. J. Clin. Microbiol. 37:681-685. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Notomi, T., Y. Ikeda, A. Okadome, and A. Nagayama. 1998. The inhibitory effect of phosphate on the ligase chain reaction used for detecting Chlamydia trachomatis. J. Clin. Pathol. 51:306-308. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Schachter, J. 2001. NAATs to diagnose Chlamydia trachomatis genital infection: a promise still unfulfilled. Expert Rev. Mol. Diagn. 1:137-144. [DOI] [PubMed] [Google Scholar]
- 25.Taylor-Robinson, D. 1996. Tests for infection with Chlamydia trachomatis. Int. J. STD AIDS 7:19-25. [DOI] [PubMed] [Google Scholar]
- 26.Verkooyen, R. P., G. T. Noordhoek, P. E. Klapper, J. Reid, J. Schirm, G. M. Cleator, M. Leven, and G. Hoddevik. 2003. Reliability of nucleic acid amplification methods for detection of Chlamydia trachomatis in urine: results of the first international collaborative quality control study among 96 laboratories. J. Clin. Microbiol. 41:3013-3016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Wang, S. P., and J. T. Grayston. 1991. Three new serovars of Chlamydia trachomatis: Da, Ia, and L2a. J. Infect. Dis. 163:403-405. [DOI] [PubMed] [Google Scholar]
- 28.Watson, E. J., A. Templeton, I. Russell, J. Paavonen, P. A. Mardh, A. Stary, and B. S. Pederson. 2002. The accuracy and efficacy of screening tests for Chlamydia trachomatis: a systematic review. J. Med. Microbiol. 51:1021-1031. [DOI] [PubMed] [Google Scholar]